View
3
Download
0
Category
Preview:
Citation preview
High Resolution Reference Solar Spectrum
for TEMPO/GEMS and beyondMina Kang 1, Myoung-Hwan Ahn 1,
Xiong Liu 2, Kelly Chance 2, Kang Sun 2, and Jhoon Kim 3
[1] Department of Atmospheric Science and Engineering,
Ewha Womans University
[2] Harvard Smithsonian center for Astrophysics
[3] Department of Atmospheric Science, Yonsei University
1
Reference Solar Spectra
2
Reference Solar Spectra
• No absolutely calibrated solar reference spectrum from single source due to limitations– Differs depending on the sources of data
(sampling, resolution, SRF and accuracy)
• High resolution reference solar spectrum is important for the calibration of the hyperspectral instrument such as TEMPO and GEMS
• As the calibration is sensitive to the choice of high resolution solar reference, we would like to improve over existing high resolution reference spectra
Reference Solar Spectra
• Reference solar spectra available for TEMPO/GEMS – Considering the radiometric, spectral accuracy, spectral sampling and
resolution for TEMPO & GEMS (high-resolution spectrometer)
Few high-resolution reference spectra are available
3
SpectrumWavelength range [nm]
Spectral resolution
[nm]
Spectral sampling
[nm]Uncertainty [%] references
NRLSSI2 115.50-99999.50 1.00 1.00 2-4 Coddington et al. (2015)
ATLAS 120.00-2400.00 0.25-0.50 0.05-0.60 2-4 for 120-2400 nm Thuillier et al. (2004)
WHI 0.10-2400.00 1.00-30.00 0.01 2-4 for 116-2400 nm Woods et al. (2008)
KNMI 235.00-520.00 0.025 0.01 <4 for 300-500 nm Dobber et al. (2008)
SAO2010 200.07-1000.99 0.04 0.01 < 5 for 300 -1000 nm Chance and Kurucz (2010)
Ftsc SAO 293.00-1626.57 0.003 Chance and Kurucz (2010)
JPL (transmittance)
380.00-2400.00 0.00014 Toon (2014)
Table. Examples of currently available reference solar spectra and their characteristics
Low resolution references
High resolution references
Reference Solar Spectra
• JPL transmittance
– Newly derived solar linelist
– Very high resolution (0.01 cm-1) with broad wavelength range (600-26316 cm-1 , 2400 -378 nm)
– not radiometrically calibrated
4
Spectral Calibration
• Sensitivity tests for GEMS wavelength calibration
5
Figure . Flow chart of spectral calibration algorithm for GEMS
− Tested for the GEMS ozone retrieval range (300 to 340 nm)
− Input perturbation
− Considered spectral parameters Reference solar spectrum SRF SNR
− Check the algorithm performance due to uncertainty of spectral parameter
Shift Squeeze FWHM
1 (%) 0.5 (%) +0.1 (nm)
Spectral Calibration
• What if we use a different reference spectrum?
6
Algorithm input Algorithm output
Shift Squeeze FWHMRef.
Solar spectrum
Truesolar
spectrumShift Squeeze FWHM
Chisquare
-0.0100 0.0050 0.7000 SAO2010 SAO2010 -0.0102 0.0050 0.6987 5.456e-6
-0.0100 0.0050 0.7000 KNMI KNMI -0.0102 0.0050 0.6989 4.887e-6
-0.0100 0.0050 0.7000 SAO2010 KNMI -0.0132 0.0050 0.6988 2.563e-4
-0.0100 0.0050 0.7000 KNMI SAO2010 -0.0068 0.0050 0.7018 2.751e-4
Synthetic (SAO)Reference (KNMI)
Synthetic (SAO)Reference (KNMI)
Before fitting After fitting
• Sensitivity tests results
− It is highly sensitive to the reference solar spectrum Due to the uncertainty of reference solar spectrum,
– Calibration performance could be considered as lower than actual performance
– The derived shifts could be different from actual wavelength variability
Minimum chi-square value increases more than an order
7
Algorithm Input Algorithm Output
Reference spectrum
Assumed measurement
Χ2
(e-6)ΔR (%)
KNMI KNMI 4.89 0.012
SAO2010 SAO2010 5.46 0.014
KNMI SAO2010 275 0.077
(ΔR is mean differences of radiance)
Spectral Calibration
Current References
• SAO2010 vs. KNMI reference spectrum
– There are non-negligible differences between the two at TEMPO/GEMS resolution (FWHM = 0.6 nm)
8
Differences between SAO2010 and KNMI (FWHM= 0.6 nm)
% differences with 0.6 nm FWHM Gaussian SRF
SAO 2010
KNMI
-----Convolved KNMI
Convolved SAO
Comparison between SAO2010 and KNMI at initial resolution and lower resolution (FWHM= 0.6 nm)
Current References
• SAO2010 vs. KNMI reference spectrum
– Differences between two are mainly caused by using of the different data sources
9
SAO2010 is not absolutely calibrated!
Wavelength range/sampling ratio/resolution (FWHM)
Requirements for Update
• Preparation of reference spectrum for TEMPO/GEMS
– TEMPO/GEMS top-level requirement
– Reference solar spectrum requirementAbsolute radiometric accuracy < 3-4% (Dobber et al.,2008)
Original spectral resolution > TEMPO/GEMS resolution
Fully Nyquist sampled (Chance et al.,2010)
10
SAO2010 and JPL (over 380 nm) are only great references for TEMPO&GEMS and beyond after calibration with low resolution (WHI below 310 nm & ATLAS over 310 nm)!!
TEMPO GEMS
Observational Range [nm] 290 -740 300 - 500
Spectral Sampling [nm] 0.2 0.2
Spectral Resolution [FWHM, nm] 0.6 0.6
Method
11
Figure. Process for updating SAO2010
• Updating SAO2010/JPL① Fitting the SAO2010/JPL with
reference (WHI/ATLAS) to obtain SRF of reference & scaling factors; same spectral feature
② Convolution of the SAO2010/JPL spectrum with the derived function ;same spectral resolution
③ Calculating the ratio between convolved spectrum with reference (R)
④ Obtaining the new convolved SAO2010/JPL applied ratio (R’) to SAO2010/JPL; new convolved SAO2010/JPL
= convolved SAO2010/JPL * R
Method
• Deriving the SRF of references based on Super-Gaussian function (Beirle et al.,2016)
12
1. 𝑆𝑝𝑒𝑐𝑡𝑟𝑎𝑙 𝑓𝑖𝑡𝑡𝑖𝑛𝑔 (OE method)
𝐼𝑟𝑒𝑓 = 𝐼𝑆𝐴𝑂𝜙𝑟𝑒𝑓𝑑λ
𝜙𝑟𝑒𝑓𝑑λ× 𝑠𝑐𝑎𝑙𝑖𝑛𝑔 𝑓𝑎𝑐𝑡𝑜𝑟𝑠
𝑺 𝒙 =𝒌
𝟐𝝕𝜞 𝟏/𝒌𝒆−𝒙𝝎
𝒌
ϖ is for the width k is for the shape of S
Result
• A significant improvement in the shorter wavelength range is evident
13
Updated SAO spectrumComparison between before and after update
(=𝑢𝑝𝑑𝑎𝑡𝑒𝑑 𝑆𝐴𝑂2010−𝑆𝐴𝑂2010
𝑢𝑝𝑑𝑎𝑡𝑒𝑑 𝑆𝐴𝑂2010∗ 100(%))
Result
• Comparison new SAO2010 with other reference solar spectra
− Increased consistency among the spectra with the decreased mean radiometric differences
14
Reference SpectrumCorrelation Coefficient
Mean RadiometricDifferences [%]
SAO2010/NRLSSI2 0.9659 1.7275
SAO2010/ATLAS3 0.9738 0.3223
SAO2010/KNMI 0.9464 1.6985
NewSAO2010/NRLSSI2 0.9754 1.4045
NewSAO2010/ATLAS3 0.9899 0.0854
New SAO2010/KNMI 0.9689 1.0593
Comparison among reference spectra for 300 -500 nm range at FWHM 1.0 nm
Result
• Calibrated JPL
– We would compare calibrated JPL with SAO2010 within oxygen air band A & B to figure out which is better
15
Discussion and Further Study
16
• Radiometric calibrated high resolution reference solar spectra (SAO2010 & JPL spectra) have been derived based on low resolution reference spectra (WHI & ATLAS) with Super-Gaussian function
• Additional optimization and detailed analysis among reference solar spectra is in progress
• We will derive high resolution solar reference spectra from 200-2400 nm at various spectral resolution (e.g., 0.01 nm, 0.01 cm-1)
Thank You
Reference Solar Spectra
18
NRLSSICoddington et al. (2015)
ATLASThuillier et al. (2004)
WHIWoods et al. (2008)
KNMIDobber et al. (2008)
SAO2010Chance & Kurucz (2010)
Different sampling, resolution, magnitude
Low resolution references
High resolution references
Result
• Updating SAO2010
19
Fitting SAO based on WHI Ratio for each iteration stepRatio btw fitted SAO & WHI
Whole Heliosphere Interval (WHI) (Mar-Apr 2008)
• obtained from simultaneous observation during solar cycle minimum conditions
• 0.1-6 nm: XPS-TIMED
– Thermosphere, Ionosphere, Mesosphere, Energetics, and Dynamics
– XUV (X-ray UV) Photometer System (XPS)
0.1-nm resolution from spectral model fitted to 7-nm broad band XPS measurement
• 6-105 nm: MEGS (rocket)-EVE
– Multiple EUV Grating Spectrograph
– EUV Variability Experiment (EVE)
0.1-nm spectral resolution measurement binned down to 0.1-nm intervals higher resolution spectrum (0.01-0.025 nm intervals) also available rocket launched on 14-Apr-08
• 105-116 nm: SEE-TIMED
– Solar EUV Experiment (SEE)
0.1-nm intervals, but 0.4-nm spectral resolution
• 116-310 nm: SOLSTICE-SORCE
– Solar Radiation and Climate Experiment (SORCE)
– Solar Stellar Irradiance Comparison Experiment
.1-nm spectral resolution measurement binned down to 0.1-nm intervals
• 310-2400 nm: SIM-SORCE
– Spectral Irradiance Monitor
low-resolution prism measurement is interpolated to 0.1-nm intervals
SIM resolution is about 1 nm at 300 nm and about 34 nm at 1200 nm
Thuillier et al 2004
• widely used for the low resolution solar reference
• 0.5-120 nm: Rocket
• 120-200 nm: SUSIM-UARS/SOLSTICE-UARS
• 200-400 nm
– SUSIM-UARS/SOLSTICE-UARS/
– SUSIM-ATLAS1-3/SSBUV-ATLAS1/
– SOLSPEC-ATLAS1-3/SOLSTICE-ATLAS1-3
• 400-870 nm: SOLSEPC-ATLAS1-3
• 870-2400 nm: SOSP-EURECA
• An empirical solar linelist has been generated by simultaneous fitting of ATMOS, MkIV, Kitt Peak, Denver U, and TCCON spectra. The telluric absorptions were fitted using the HITRAN linelistand any remaining, airmass-independent absorptions were attributed to the sun. Together with a simple lineshape function, this linelistallows the computation of a solar pseudo-transmittance spectra, for disk-center, disk-integrated, or intermediate cases. These spectra have been validated using MkIV, Kitt Peak, ACE, and GOSAT spectra.
22
Solar Update
• SAO 96– 230-800 nm, 0.01 nm resolution/0.02 FWHM
– Based on ground-based measurements (Kurucz 1984) & balloon measurements (Hall and Anderson 1991) Converted to vacuum wavelengths.
Ground data was recalibrated using 4 O2 lines (accuracy of better than 0.001 nm)
Balloon data were recalibrated in wavelength using 20 selected atomic reference lines. (accuracy of 0.003 nm)
Spectra were resampled at even 0.01 nm increments, employing a triangular filter of 0.02 nm FWHM and linearly merged over the 300-305 nm
Spectrum
Wavelength/Sampling (nm)
Resolution (nm)
Sources Accuracy
Kitt Peak
293.09-1627.02 vacuum nm derived from the Fourier transform spectroscopy measurements of the Sun
0.0003 – 0.001
0.0005
Kurucz R. L., I. Furenlid, J. Brault, and L. Testerman, SolarFlux Atlas from 296 to 1300 nm (National Solar Observatory,Sunspot, New Mexico, 1984) 240 pp.
http://kurucz.harvard.edu/sun.html
Hall and Anderson
200-310 0.01 measured near 40 km in the stratosphere in 1978 and 1983, have been combined and extrapolated to zero optical depth to provide a reference spectrum in 0.1-Å
0.025
Hall L. A. and G. P. Anderson (1991), High-resolution solar spectrumbetween 200 and 3100 Å,” J. Geophys. Res. 96, 12, 927–12,931
within 0.04 Å.
Solar Update
• SAO produce the calibrated solar spectrum (235-1100 nm) as part of OMPS instrument and algorithm development – SOLSTICE and SUSIM (Woods et al., 1996)+
LOWTRAN/MODTRAN– Linearly merged 405.6-410.6 nm
• OMI spectrum (Dobber et al., 2006)– Kitt peak (Kurucz 1984) and Hall & Anderson 1991– Use combined low resolution spectra
SUSIM from UARS (Floyed et al., 2003) up to 410 nm, 0.15 nm resolution
– Triangular slit
Balloon spectrum for SCIAMACHY validation (Gurlit et al., 2005) from 400 nm
– Square shaped slit function
Solar Update
• SAO 2010– Use updated kitt peak (agreement with Thuillier et
al., ; better than 1% at 20 nm spectral resolution)Large scale features by O3, O2-O2 were computed and
divided out.The computed line spectrum were adjusted for an match to
the observed spectraArtifacts from wavelength mismatches, deep lines, etc, were
removed by hand.
– Updated kitt peak data and H&A spectra are each convolved to a Gaussian FWHM of 0.04 nm and sampled to 0.01 nm
– Linearly merged to 300 to 305 nm
Solar Update
• Compare SAO 2010 & OMI
Range (nm)
Sampling (nm)
Resolution (nm)
High resolution
Low resolution issues
SAO 2010200.07-1000.99
0.01 0.04Kitt peak (2005)Hall & Anderson
Radiometriccalibration
OMI 250-550 0.01 0.025Kitt peak (1984)Hall & Anderson
(SAO 96)
SUSIM (~410 nm)(Floyd et al 2003)
Balloon (400 nm ~)(Gurlit et al 2005)
Under-sampling
Updated SAO
200.07-1000.99
0.01 0.04 SAO 2010
WHI (~310 nm)(Woods et al 2009)ATLAS1&3 (300
nm~)(Thuillier et al 2004)
FWHM
Solar Update
• Allow overlapping between adjacent windows.
– For example, if each window is 10 nm, then fit 292-302 nm, 300-310 nm, 308-318 nm and only apply the results to non-overlapped regions, i.e., results from fitting 300-310 nm to 302-308 nm.
FWHM
FWHMFWHM
292 302
300 310
308 318
312 322
𝐼𝑊𝐻𝐼 = 𝐼𝑆𝐴𝑂 × 𝜙𝑊𝐻𝐼𝑑𝜙
𝜙𝑊𝐻𝐼𝑑𝜙× (𝑎𝑜 + 𝑎1 𝜆 − 𝜆𝑎𝑣𝑔 + 𝑎2 𝜆 − 𝜆𝑎𝑣𝑔
2+ 𝑎3 𝜆 − 𝜆𝑎𝑣𝑔
3+ 𝑎4 𝜆 − 𝜆𝑎𝑣𝑔
4+ 𝑎5 𝜆 − 𝜆𝑎𝑣𝑔
5
• Convolved SAO based on ratio after iteration Nearly same as original WHI spectral region
• Convolved SAO based on obtained scaling factors from fitting Quite different with original WHI the fitted SAO
includes the broad variation of the correction from the scaling parameters
considered ratio
Current References
244.92285
242.96128
240.80794
Calculate TSI for 350-500 nm
Current References
245.16330
242.68037
245.16014
Similar with Gurlit et al
Similar with Kurucz
Recommended